There has conventionally been known a technique of implementing phase-difference focus detection AF (Auto Focus) (phase difference AF) by giving a phase-difference focus detection function to an image sensor without using a dedicated AF sensor.
For example, in Japanese Patent Laid-Open No. 2001-305415, some pixels of an image sensor are formed as focus-detecting pixels by giving a pupil division function to them. The pupil division function is provided by dividing the light receiving portion of a pixel into two. Outputs from the two light receiving portions are used as phase-difference focus detection signals, and the sum of them is used as a pixel signal.
In Japanese Patent Laid-Open No. 2000-156823 or 2008-134389, the pupil division function is given to some pixels of an image sensor by forming a rectangular opening in a light shielding layer between a microlens and a photoelectric conversion unit so that the opening is decentered from the optical axis of the microlens. These pixels having the pupil division function are arranged as focus-detecting pixels at predetermined intervals to perform phase-difference focus detection AF. Image signals at portions where the focus-detecting pixels are arranged are generated from image signals from peripheral pixels.
However, according to the technique disclosed in Japanese Patent Laid-Open No. 2001-305415, the image sensors need to be configured so that the light receiving portion of a pixel to be given the pupil division function is divided into two areas, and photoelectrically converted signals in the divided areas can be independently read out. This complicates the structure of the image sensor, which is disadvantageous in decreasing the pixel pitch. An area between the divided areas becomes a blind zone, so the sensitivity of the image sensor may decrease.
According to the techniques disclosed in Japanese Patent Laid-Open Nos. 2000-156823 and 2008-134389, a light shielding layer for pupil division needs to be arranged. However, arranging the pupil division-specific light shielding layer complicates the structure of the image sensor. Especially these days, CMOS image sensors (solid-state image sensors using a complimentary metal oxide semiconductor) have widely prevailed. The CMOS image sensor requires three electrode layers (interconnection layers) or so between a microlens and a photoelectric conversion unit. When the technique disclosed in Japanese Patent Laid-Open No. 2000-156823 or 2008-134389 is applied to the CMOS image sensor, the light shielding layer is added to the interconnection layers. The distance between the microlens and the photoelectric conversion unit will increase, decreasing the light receiving efficiency. In general, the focus position of the microlens is set near the surface of the photoelectric conversion unit. If the light shielding layer for pupil division is arranged immediately below the microlens, no desired pupil division performance can be obtained, and the focus detection precision may decrease.